The histomorphological features of ovarian and colorectal adenocarcinomas overlap, at times. In certain situations, the ovarian metastasis form a colorectal adenocarcinoma might be the primary presentation or a colorectal tumor might actually be a metastatic adenocarcinoma from the ovary [3, 6]. Identification of an exact primary tumor type i.e. either ovarian or colorectal, in this scenario becomes vital for an appropriate management. Apart from certain clinicopathological features, lately, differential expression of cytokeratins 7 & 20 has been found to be useful in resolving these dilemmas [9–11]. The present study was based on clinicopathological features and identification of value of differential expression of CK 7 and 20 by immunohistochemistry, in cases of ovarian involvement by colorectal adenocarcinomas and. in cases of colorectal involvement by ovarian adenocarcinomas.
Among the already described clinical features, primary ovarian adenocarcinomas are usually know to occur at a relatively younger age group, as was seen in our study (median age = 49 years), in contrast to ovarian metastatic deposits from colorectal adenocarcinomas that were seen in an older age group (mean age = 55 years), as noted in an earlier study . In our series, involvement of the ovary and colon in cases of colorectal metastasis was more commonly synchronous, in contrast to the findings of Lewis et al , wherein the ovarian lesions were identified prior to the colonic primary in 32% cases. This might be due to the fact that that ours being a referral centre, majority of the patients present at an advanced stage disease, along with existing lack of routine gynaecologic or gastrointestinal (GI) screening.
The differences in the pre and perioperative characteristics of primary and secondary ovarian tumors have been studied by Antila et al , who found that tumors that were relatively smaller -10 cms sized; solid in nature; presenting with ascites; in association with raised serum CEA & tumor-associated trypsin (TAT-1) inhibitor levels, were predominantly metastatic. Metastatic tumors are reported to be commonly bilateral [1, 6, 11]. However, we observed 66% of metastatic colorectal adenocarcinomas in the ovaries as unilateral masses, in as noted by Daya et al . In their study, Lewis et al  have also reinforced this fact, with a suggestion of consideration of metastatic tumor in cases of younger patients with large sized tumors. Wherever the gross findings were available in our study, all such cases had a capsular breach and were both solid and cystic in maximum cases. The mean ovarian tumor size could not be ascertained in view of referral cases being submitted with limited details, adding to the diagnostic challenge. In addition, 90% of such cases displayed raised serum CEA levels, wherever available.
On histology, 'garland-like' necrosis with desmoplasia were the most valuable clues in diagnosing a metastatic colorectal carcinoma, as noted in earlier studies [2, 5, 6]. However, we did not observe stromal luteinization in any of our cases as noted by Mc Cluggage et al  in their cases. Metastatic mucinous adenocarcinomas can closely mimic primary ovarian, although, the latter, is invariably known to display areas resembling borderline primary ovarian neoplasms. However, this feature can be seen in a metastatic colorectal adenocarcinoma . Both our cases of this subtype had histological, as well as gross evidence of surface involvement, with discrete features of a colonic adenocarcinoma, including 'garland-like' tumor necrosis. Presence of lymphovascular emboli favoured a metastatic carcinoma over a primary. Metastatic colorectal adenocarcinomas invariably mimic an endometrioid type of ovarian adenocarcinomas. Identification of foci of squamous differentiation, endometriosis or adenofibromatosis are pointers towards an ovarian adenocarcinoma. Despite these clinicopathological features, difficulty exists, wherein IHC is valuable in forming a more objective diagnosis .
Lately, cytokeratins 7 and 20 have been found to be useful in providing a direction towards exact primary in cases of metastasis with unknown primary (MUP) . Cytokeratin 7 (CK7) is a basic (type II) cytokeratins, found in human ductal, glandular and transitional epithelia, while CK 20 it is normally expressed in gastrointestinal epithelium, urothelium and Merkel cells [13, 14]. As described lately, these specific cytokeratins, with additional markers like CEA and CA125, forming an optimal IHC panel, have been documented to be helpful in resolving these dilemmatic situations of ovarian involvement by colorectal carcinoma and vice versa. The colonic tumors usually express CK 20 and CEA diffusely and are negative for CK7 and CA-125, the latter two being more frequently expressed by ovarian adenocarcinomas in this clinical context. This pattern of IHC expression helped in ascertaining metastatic colorectal adenocarcinoma in the ovary in 8/9 cases in our study. The remaining 1 case was spared with CK 20 expression since it had identical histomorphological of the colorectal adenocarcinomas that displayed a transition from the dysplastic colorectal mucosa. CK7 was focally positive in 3 such tumors, as noted earlier.  Moreover, CK 20 and CEA positivity can be observed in mucinous ovarian neoplasms. However, the expression is focal, unlike diffuse expression noted in lower GI adenocarcinomas [6, 7, 16]. In our limited number of cases in this clinical context, we did not find any pure mucinous ovarian tumor, although, focal mucinous differentiation was noted in a single case. Serum CEA levels were marginally elevated, but IHC for the same was negative. Lately, CDX2 and villin have been found to be 100% positive in cases of metastatic colorectal adenocarcinomas . However, CDX2 can be seen in mucinous ovarian adenocarcinoma.
Eleven cases in our series were identified as primary ovarian adenocarcinomas, metastasizing to the colorectum. 90% of these cases were metachronous, with a preceding ovarian tumor. On gross examination, these were included predominantly in the bilateral ovarian tumors, with capsular breach and solid and cystic areas. On morphology, metastatic tumor deposits in the colorectum retained the morphology of ovarian papillary serous cystadeocarcinoma with foci of psammomatous calcification in 6/11 cases. Absence of 'garland-like' tumor necrosis, desmoplasia and LVI were pointers towards an ovarian primary, apart from the clinical context of a preceding ovarian tumor in 81.8% such cases. Wherever the colonic resection specimens were available, the tumor was predominantly submucosal or serosal, with absence of surface dysplasia, ruling out a colonic primary. In view of a definite clinical diagnosis with identical histomorphological features of the primary ovarian adenocarcinoma, IHC was restricted to 6 out of 11 such cases. In these cases the IHC pattern included CK7 and CA-125 positivity, with CEA and CK 20 negativity . Four of these were metachronous (3 treated cases, post CT). Zighelboim et al  described a single case of atypical sigmoid metastasis from a high-grade ovarian adenocarcinoma, using differential expression of CK7 & 20. The plausible explanation of colorectal involvement in ovarian adenocarcinomas is through intraperitonel seedling. Hematogenous spread occurs in advanced peritoneal disease. In a substantial number of autopsy cases of advanced ovarian cancers, Rose et al  identified peritoneal involvement in 83–100% cases, with large intestinal involvement in 50–60% cases. An objective confirmation with specific cytokeratins in our 6/11 such unusual cases series has not been described earlier. In most of such cases, serosa was found to be initially affected and the remaining wall, thereafter. In their study, O' Hanlan et al  hypothesized that ovarian metastases to the gastrointestinal tract might mimic the colonic pattern of dissemination. In this setting, the invaded lymph channels at the submucosa of the bowel enter and replace mesenteric lymph nodes, forming large deposits of carcinoma in the paracolic and intermediate mesenteric groups. They concluded that a longitudinal negative margin of 2 – 5 cm in the resected bowel along with a wedge resection of mesentery, including paracolic and intermediate-level nodes might be indicated to achieve optimal debulking of gastrointestinal metastases from ovarian carcinomas.
Reardon et al  described an ovarian papillary serous cystadenocarcinoma mimicking recurrence at colorectal anastomosis in a previously diagnosed case of a colorectal adenocarcinoma, utilizing progesterone receptor (PR) expression. The possibility of double cancers in our study was ruled, based on the aforementioned clinicopathological features. Nonetheless there were challenging cases. A case of an ovarian tumor, on biopsy, showed an adenocarcinoma with both, 'garland-like' necrosis and psammomatous calcification. Presence of psammomatous calcification has been rarely documented in colorectal adenocarcinomas . On IHC, diffuse CK20 and CEA positivity; with CK7 and CA-125 negativity reinforced a colonic primary that was further confirmed by presence of a rectal tumor. Another case presented with a rectal nodule, which was found to be a serosal deposit of a poorly differentiated adenocarcinoma. Imaging revealed an adnexal mass. Histopathologically, unremarkable rectal mucosa with CK7 positivity and CK 20 negativity in the tumor cells helped in ruling out a colonic primary. CK 7 is also known to be expressed in breast and the upper aerodigestive tract . Additionally CA-125 positivity, along with elevated serum tumor markers levels of CA-125, confirmed an ovarian origin of the tumor.